bigint search

[kevindawson]

when searching for bigint the expected first match should be -> F/FL/FLORA/bignum-0.32.tar.gz looks like we have another search index issue.

If we look at -> https://metacpan.org/author/FLORA there is no bignum

but here it is on CPAN http://search.cpan.org/~flora/bignum-0.32/lib/bigint.pm

oalders new issue as per your instruction, sorry for the delay in doing so.

[wolfsage]

Perhaps this is because the latest release of bignum (which contains bigint) is an unauthorized release by a different author:

https://metacpan.org/release/PJACKLAM/bignum-0.37

This is also screwing up FLORA's module list page. It should still show .32 for bignum in his list:

https://metacpan.org/author/FLORA

But does not since bignum-0.37 belongs to PJACKLAM. (Shouldn't bignum-0.37 show a big "UNAUTHORIZED RELEASE" warning?)

[rwstauner]

Likely adding to the confusion is that, as of Sun, 11 May 2014 10:41:02 GMT, 02packages lists two different versions of bignum:

bigint                             0.32  F/FL/FLORA/bignum-0.32.tar.gz
bignum                             0.32  F/FL/FLORA/bignum-0.32.tar.gz
bigrat                             0.32  F/FL/FLORA/bignum-0.32.tar.gz
Math::BigFloat::Trace              0.36  P/PJ/PJACKLAM/bignum-0.37.tar.gz
Math::BigInt::Trace                0.36  P/PJ/PJACKLAM/bignum-0.37.tar.gz

According to 06perms PJACKLAM is authorized for the Math::.+ modules in the dist, but not the ^big*ones.

The dist is unauthorized according to the API:

*$ curl api.metacpan.org/release/PJACKLAM/bignum-0.37?fields=authorized,status
{
   "authorized" : "false",
   "status" : "latest"
}

So there must still be a bug about marking the dist as latest because it has some modules that are in 02packages but it is partially unauthorized.

There also doesn't appear to be code in the web ui about displaying authorized status.

[rwstauner]

I updated the UI to mark the release as unauthorized, but the dist is still incorrectly marked as latest. I made an issue on the api for that (cpan-api/cpan-api#315).

[shadowcat-mst]

The heuristic I'd suggest is to take the 'king module' of a dist - i.e. if somebody ships DBIx-Class and they're authorized for DBIx::Class, they're probably supposed to be able to ship the whole thing. Noting 'permission issue' or something-or-other for partially authorized dists is good, but it's common enough for a couple of modules deep in a namespace to get missed when distributing co-maint that you don't want to slap the big UNAUTHORIZED label on them just for that and you certainly don't want to exclude them from being the latest

[rwstauner]

That's not a bad idea. This bignum case would still show RAFL's as latest (which would agree with PAUSE), but it wouldn't incorrectly break the majority of cases (where installing by the main module would still get you the newest one).